Christmas in Aruba · Seasonal E-commerce
Spent $922 on ads. Booked $3,670 in orders.
3.98×
return on ad spend, 28 July to 3 September
A Christmas retailer on a Caribbean island sells a year of trees in one pre-order window. We built the ad engine that finds its buyers, then stopped paying for the ads that filled carts without closing them.
One island, one selling window, no room to waste a dollar
Christmas in Aruba sells pre-lit artificial trees, wreaths and garlands to a market of about 106,000 people. There is no winter on the island. Christmas is a cultural season, and the entire business runs on it.
The model is pre-order. Customers reserve between June and September, stock is imported against confirmed demand, and every order is delivered before the first of November. There is no second quarter to make up a bad one.
The account had run a single test campaign: $128 over ten days, optimized for reach, on a pixel that reported page views and nothing else. It spent, it got clicks, and nobody could say whether one sale came from it.
Make the pixel tell the truth, then only pay for what closes
We started with measurement. The store’s pixel was verified event by event against the live site, by loading the store and watching which events fired: page view, product view, add to cart, checkout, purchase. Until the purchase event reports back, every decision after it is a guess.
The campaign optimizes for add to cart. That is deliberate. A market this small does not produce enough weekly purchases for the algorithm to learn from, so add to cart gives it enough signal to work with. We grade the ads on what happens after the cart.
Then we read the account one ad at a time. The ad filling the most carts was returning fifty-eight cents on the dollar. Four ads were defunded and their share went to the two that closed. The daily budget never moved.
The results
A small budget, spent only where it came back
-
$922 to $3,670
ad spend and booked pre-order revenue, 28 July to 3 September
-
3.98×
return on ad spend across the full window
-
13 orders
at $70.93 of ad cost each, average order value $282
-
210
add to carts and 137 checkouts started, on $25 a day
Return on ad spend by ad, pulled from the Meta API on 10 September. Across every ad in the account the return was 3.98×.
The account average is the least interesting number here. Inside it, one ad returned fifty-eight cents for every dollar it spent while another returned $7.79. Both ran at the same time, to the same island, out of the same budget. The ad that filled the most carts was the one losing money.
Revenue here is booked pre-order value, delivered in November. On a budget this size the whole job is finding the ad that closes and cutting the one that only looks busy.
contact us
See where your store stands, free
The honest way to know whether your marketing is pulling its weight is to measure it. Start with the Visibility Benchmark: one field, your website, and a straight read on where you stand in search and inside the answers buyers now get from AI. Then we'll talk about the work.